Hardware-bound biometric confirmation. Single-use access token issued on success.
No identity is disclosed. No behavioral data is collected. Only proof of physical presence is verified.
Human presence verified.
A cryptographic proof of a physically present human has been issued.
By burning one Presence Credit you have opened the first human-attested website in the history of the internet.
This moment demonstrates the core constraint of the Human Presence Protocol: presence is no longer a policy claim. It is a cryptographic, hardware-enforced property.
Welcome to the attested internet.
All attestations are pseudonymous.
No identity disclosure. No data collection.
Patent pending · USPTO Customer No. 224891
HPP is a constraint-based attestation protocol that produces cryptographically verifiable evidence of human presence without requiring identity disclosure. It enforces thermodynamic cost symmetry between humans and bots: the human body is the non-parallelizable substrate; no software substitution is possible at the hardware attestation layer.
"Every machine proof imposes an energy tax; HPP reads a signal existence is already emitting. The human body is the proof. Existence is the work. Presence is the consensus."
42 provisional patent families · Filed February 14 – March 28, 2026 · 1,429 claims · 32 primitives · 15 invariants · 48 capability enablers · NP Deadline: February 14, 2027
10-condition Srinivasan map. HPP as architecturally enforceable antitrust remedy. FTC / DOJ / AG entry point.
For Counsel RAD DocumentationUSPTO filing dates vs. KGM verdict. Six KGM features mapped to HPP applications. MDL 3047 bellwether context.
For Academics Formal ModelThree formal notions + one physical assumption. Adversarial games. Necessity theorem. Dependency chain. USENIX NPHT submission.
HPP is a constraint-based attestation protocol. It enforces thermodynamic cost symmetry between humans and bots — not by detecting bot behavior, but by requiring physical costs that only a human body can produce. The human body is the non-parallelizable substrate; NPHT formalizes this. Architecture cannot be reversed by a platform decision. Presence Credits are a consequence of Biometric Burn, not a standalone economic invention. HPP does not require identity disclosure; all attestations are pseudonymous and bound only to hardware, not identity.
HPP v1.2 §1.1 defines a notion as "a formally named, quantified property that parameterizes the adversarial model." Under this definition, the construction rests on three formal security notions (HPP-PRES, NPHT, Biometric Burn) and one physical assumption (H-Constant) that functions as a model-level notion. All four are required. Removing any one collapses the construction. The H-Constant is not proven within the model — it is assumed, in the same way that hardness assumptions (LWE, DDH, random oracle) are assumed in cryptography. All security claims that depend on it are assumption-conditional.
Notions are properties; primitives are mechanisms; assumptions are premises.
HPP is: a constraint-based attestation protocol · pseudonymous by design · hardware-bound, not identity-bound. HPP is not: biometrics-as-identity · surveillance · behavioral detection · a social credit system · a token economy.
| Element | Category | Role | Proven? |
|---|---|---|---|
| HPP-PRES | Security notion | Unforgeability | Proven conditional on assumptions |
| NPHT | Throughput notion | Non-parallelizability | Proven conditional on assumptions |
| Biometric Burn | Closure notion | State transition validity | Proven conditional on assumptions |
| H-Constant | Physical assumption | Energy lower bound | Assumed, not proven |
The adversary is any PPT attacker with arbitrary compute, arbitrary software control, and no access to hardware-bound biometric liveness channels.
The core cryptographic security notion. HPP-PRES defines the adversarial unforgeability game: a probabilistic polynomial-time adversary A with query access (q_n nonce queries, q_b biometric queries, q_h hardware queries) cannot produce a valid presence attestation without satisfying the physical hardware-bound biometric requirements. Formally: Pr[A wins HPP-PRES] ≤ Adv_sig(λ,q,t) + q_n/2^λ + q_b·ε_live + q_h·ε_hw. This bound is assumption-conditional on the H-Constant holding. Revised following external review; see Section 6.
The adversarial throughput notion. NPHT formalizes the non-parallelizability of valid attestation sequences: the throughput of valid attestations is bounded above by the physical capacity of a single human being operating a hardware-bound device over a real-time Attestation Window. No software multiplier can substitute for the biological substrate. Bot farms cannot scale valid attestation output because the cost is per-human per window, not per compute cycle. NPHT is the formal statement of why the human body is the non-parallelizable substrate. Assumption-conditional on the H-Constant. NPHT is not a performance metric; it is a security bound on adversarial throughput.
The H-Constant (≈8 joules per attestation) is a physical lower-bound assumption — not a security notion proven within the model, but a quantified, named, load-bearing assumption that parameterizes the adversarial model. It functions as a model-level notion in the sense of v1.2 §1.1: a formally named, quantified property on which the adversarial game depends. It is assumed in the same way that hardness assumptions (LWE, random oracle) are assumed in cryptography — not derived, but stated as a condition on which all dependent security claims are conditional. The H-Constant cannot be set to zero by a software update; it is grounded in the physical energy required to produce a valid Biometric Liveness Gate signal on a Hardware-Bound Key device. The H-Constant is environment-dependent; its value is not universal but must be bounded below by a positive constant in any valid instantiation (v1.2 §1.4). It does not assert a specific numeric value for any hardware platform; it asserts only that a positive lower bound exists.
The consumption-closure notion. Biometric Burn is a formal state predicate: a valid attestation event transitions the attestation state of a hardware-bound device from an open-window state to a closed-window state, consuming a Biometric Liveness Gate signal that cannot be replayed or delegated. Presence Credits are the output of this state transition — a consequence of Biometric Burn, not a standalone economic invention. The notion closes the construction: no attestation is valid unless the hardware-bound liveness gate has been consumed in that Attestation Window. Biometric Burn is not a rate limit; it is a state-predicate closure condition (v1.2 Definition 6).
Each element is required. The chain flows left to right: the physical assumption grounds the throughput notion, which grounds unforgeability, which closes via state predicate into the Presence Credit output. The dependency chain is directional; later notions cannot substitute for earlier ones.
Presence Credits are not a currency, token, or transferable asset; they are a per-window attestation output.
The Minimal Constraint Set consists of three formal notions (HPP-PRES, NPHT, Biometric Burn) and one physical assumption (H-Constant). All four are required. Removing any one collapses the construction under the adversarial model of v1.2. The Minimal Constraint Set is not tunable; all four elements must hold simultaneously.
Server-time-authoritative gating and Hardware-Bound Keys are necessary implementation prerequisites — required for any valid instantiation — but they are not themselves the formal constraint set. They are conditions on which the notions depend, not the notions themselves.
Attestation Windows are set by server time, not client clock, preventing replay and pre-computation attacks. Required for NPHT and HPP-PRES to hold. An Attestation Window is a server-defined real-time interval during which a biometric liveness event must occur for an attestation to be valid.
Attestation is cryptographically bound to a specific physical device via Secure Enclave or equivalent TEE. Required for the H-Constant assumption and Biometric Burn notion to hold. Hardware-Bound Keys are non-exportable keys generated inside a Secure Enclave or equivalent TEE; they cannot be cloned, copied, or transferred.
Primitives are algorithmic or architectural building blocks — they are distinct from notions. Notions are formal security properties; primitives are the specific mechanisms that implement them. Layer 8: P1–P15 (Human Presence Protocol) · Layer 9: G1–G9 (Constitutional Governance, open-sourced CC) · Layer 10: E1–E8 (Economic Framework). The 48 Capability Enablers (CE1–CE48) span all rings and express how primitives combine into deployable functions.
Hardware Attestation. Secure Enclave / TEE binding. Pro-A (App. No. 63/983,106) and Pro-K (App. No. 63/983,136). Filed February 14, 2026. Pro-A and Pro-K are the primary RAD anchors for the portfolio — 39 days before the KGM verdict.
Enforcement Economics. H-Constant physical lower-bound notion. Presence Credits as consequence of Biometric Burn. Non-transferability. Pro-L (App. No. 63/985,404) filed February 18, 2026. Pro-L is a RAD anchor for enforcement economics — 35 days before verdict.
Verticals. Age verification (Pro-P, RAD anchor, February 18, 2026). Child safety (Pro-M, March 27, 2026). Active/Passive presence tiers (Pro-Q, App. No. 64/018,841, March 27, 2026). Pro-Q is a post-verdict enforcement architecture — not a RAD anchor.
15 System Invariants. INV-01 through INV-15. Top-Level Domain namespace. INV-15 (Endpoint-Bound Attestation, App. No. 64/018,908) confirmed March 27, 2026.
Economic primitives. ECO-A through ECO-H define the Presence Credit supply mechanics, settlement architecture, and cross-platform interoperability structures. These are architectural primitives — not market mandates or policy prescriptions.
L8: Human Presence Protocol · L9: Constitutional Governance (open-sourced CC) · L10: Economic Framework · L11: Civilization Design
HPP does not require identity disclosure; all attestations are pseudonymous. The protocol answers one question only: is this signal consistent with a physically present human being? The following misinterpretations are common and incorrect.
HPP uses a biometric signal to confirm liveness — the physical presence of a human being in an Attestation Window. It does not use biometrics to identify who that human is. P12 (Commitment-Based Presence Proof) enables zero-knowledge verification: a platform receives proof of presence, not identity.
HPP does not record, transmit, or aggregate user behavior. The attestation is a point-in-time cryptographic proof. It says "a human was present at this device at this window." It does not say what that human did, read, said, or bought.
HPP is constraint-based, not detection-based. It does not analyze patterns of behavior to infer whether a user is a bot. It enforces a physical cost that bots cannot pay. The distinction is foundational: detection degrades as AI improves; constraint does not.
Presence Credits are non-transferable, non-tradeable, and not linked to behavior, content, or social standing. They record only that a human was present. They expire daily. They cannot be accumulated across identities or used for scoring.
Presence Credits are not a cryptocurrency, a blockchain token, or a monetizable asset. They are an architectural state output of Biometric Burn — a formal closure predicate. Their function is attestation validity, not exchange value.
HPP is an architectural framework presented for evaluation. It does not assert that any regulator, court, or platform has adopted or is required to adopt it. All materials on this site are provided for informational and technical evaluation purposes only.
HPP is designed to defeat the following adversarial classes under the v1.2 adversarial model. All security claims are assumption-conditional on the H-Constant.
| Adversary Class | Attack Vector | HPP Defense | Relevant Notion |
|---|---|---|---|
| Bot farm operator | Parallel software agents generating fake attestations at scale | H-Constant: each valid attestation requires ≈8 J of physical energy from a hardware-bound device. Bot-scale deployment is thermodynamically prohibitive. | H-Constant (assumption) · NPHT |
| Replay attacker | Capturing and replaying valid attestations from legitimate users | P13 (Replay-Resistant Attestation Windows): each attestation is tied to a server-time-authoritative window and is invalid outside it. Replayed attestations fail window validation. | HPP-PRES · P13 |
| Key exfiltration attacker | Exporting Hardware-Bound Keys to produce attestations on non-target hardware | Hardware-Bound Keys are generated inside Secure Enclave / TEE and are non-exportable by design. Exfiltration is outside the threat model — assumed infeasible under the hardware attestation prerequisite. | HPP-PRES · P2 |
| Sybil attacker | Creating multiple virtual identities to accumulate Presence Credits above per-human biological capacity | P14 (Anti-Sybil Presence Accumulation): Presence Credits are bounded by P6 (Single Primary Device) and Biometric Burn's per-window consumption. Multi-device accumulation requires physical multiplication of human actors. | Biometric Burn · P14 · P6 |
| Delegation attacker | Delegating biometric attestation to another human to produce credits on behalf of a third party | Presence Credits are non-transferable (P5). Credits earned on device A by human B are valid only for device A in the window they were earned. Transfer is architecturally blocked, not policy-restricted. | Biometric Burn · P5 |
| AI liveness spoofing | Using generative AI to produce synthetic biometric signals that satisfy the Biometric Liveness Gate | ε_live (Biometric Liveness Gate) is a formal parameter. Concrete instantiation determines the bound. This is an open research problem (see Section 6.4). HPP does not assert unconditional resistance to advanced AI liveness spoofing; it asserts it is assumption-conditional on ε_live. | HPP-PRES (ε_live term) |
USPTO filing dates are documentary evidence. They establish that HPP was architected, formalized, and filed prior to judicial findings of design defect. This is the foundation of the RAD record.
Six interaction design features at issue: infinite scroll, autoplay, notifications, beauty filters, like counts, algorithmic amplification. HPP not yet filed.
Pro-A through Pro-K. Ring 1 (hardware-bound biometric attestation), Ring 2 (enforcement economics), Ring 3 (interoperability). App. Nos. 63/983,106–63/983,136. Pro-A and Pro-K are the primary RAD anchors for the KGM feature set. This wave establishes the portfolio's priority date.
Pro-L: H-Constant Enforcement (App. No. 63/985,404) — physical lower-bound notion formalized. Pro-P: Age Verification (App. No. 63/985,418) — privacy-preserving age gate. Pro-L is a RAD anchor for enforcement economics. Pro-P is the RAD anchor for the beauty filter KGM feature.
Los Angeles Superior Court jury, JCCP 5255. 40+ hours deliberation over nine days. Meta: 70% ($4.2M). Google/YouTube: 30% ($1.8M). Section 230 rejected for design claims. 2,407 cases pending in federal MDL 3047. TikTok and Snap both settled pre-trial.
Pro-Q (Active-Passive Presence Tiers, App. No. 64/018,841), Pro-M (Child Safety, App. No. 64/018,735), INV-01 through INV-15 (TLD Namespace layer), ECO-A through ECO-H (economic layer). Portfolio complete: 42 families, 1,429 claims. Pro-Q and the INV layer are post-verdict enforcement architectures — they are not RAD anchors.
Breathitt County Board of Education (KY), lead bellwether. Motley Rice trial counsel. All four defendants named. Six plaintiff experts cleared over Daubert challenges. RAD documentation must be in counsel's hands before this date.
Statutory deadline to preserve February 14, 2026 priority dates. Hard legal deadline — not a negotiating posture.
RAD anchors are pre-verdict filings. Pro-Q and post-verdict filings are enforcement architectures — they document the design space but are not the primary RAD evidence. This distinction matters for expert witness framing under the RAD standard.
| KGM Feature Found Defective | Primary RAD Anchor | App. No. | Filing Date | Status |
|---|---|---|---|---|
| Notifications (engagement-maximizing) | Pro-A — Hardware-Bound Attestation Root | 63/983,106 | 02/14/2026 | 39 days pre-verdict |
| Like Counts (social validation) | Pro-A — Hardware-Bound Attestation Root | 63/983,106 | 02/14/2026 | 39 days pre-verdict |
| Algorithmic Amplification | Pro-A + Pro-K — Attestation Root + Enforcement Architecture | 63/983,106 / 63/983,136 | 02/14/2026 | 39 days pre-verdict · Pro-Q is enforcement, not RAD anchor |
| Beauty Filters | Pro-P — Age Verification Gate | 63/985,418 | 02/18/2026 | 35 days pre-verdict |
| Infinite Scroll | Pro-Q — Active/Passive Presence Tiers (enforcement architecture) | 64/018,841 | 03/27/2026 | Post-verdict enforcement architecture |
| Autoplay | Pro-Q — Active/Passive Presence Tiers (enforcement architecture) | 64/018,841 | 03/27/2026 | Post-verdict enforcement architecture |
Srinivasan (2019) documents ten structural mechanisms enabling surveillance capitalism. HPP is presented here as an architectural framework for evaluation — not as a legal conclusion or a claim that HPP has been or will be adopted as a remedy. Every behavioral commitment made by the platforms documented in the Srinivasan record was reversed: Beacon (2007), social plugins (2010–2014), Atlas (2014). Architecture cannot be reversed by a platform decision.
| # | Srinivasan Structural Condition | HPP Architectural Response | Primary Primitive |
|---|---|---|---|
| 1 | Surveillance is costless to the platform | H-Constant (≈8 joules/attestation) imposes a formal thermodynamic cost floor. Surveillance-without-attestation costs the bot operator the same as legitimate presence — asymptotically prohibitive at scale. | Pro-L / P2 |
| 2 | Users have no counter-leverage | Presence Credits (P5) are non-transferable, hardware-bound, and expire daily. They are earned by human biological signal — not purchased, rented, or delegated. The human is the leverage. | P5 / P4 |
| 3 | Cookies enable cross-site tracking | P13 (Replay-Resistant Attestation Windows) ties each attestation to a server-time-authoritative window. Cross-site tracking requires valid presence per site — bot-scale deployment cannot satisfy this without physical human multiplication. | P13 / P1 |
| 4 | Facebook knows real user identities | P12 (Commitment-Based Presence Proof) enables zero-knowledge presence verification — the platform receives proof of human presence without receiving identity. Privacy-preserving by construction. | P12 / P10 |
| 5 | Third parties installed Facebook code | P10 (API-Mediated Verification) and P9 (Trust Bridging Assertion) replace code-injection surveillance with a standardized, auditable attestation API. Third parties receive presence signals, not surveillance infrastructure. | P9 / P10 |
| 6 | User voting process was abolished | G1–G9 (Constitutional Governance Layer, Layer 9) formalizes governance as a protocol primitive — open-sourced under Creative Commons. Governance rules are embedded in the architecture, not subject to platform policy reversal. | G1–G9 |
| 7 | Do Not Track was ignored | P14 (Anti-Sybil Presence Accumulation) creates a structural, not a voluntary, exclusion mechanism. Bot traffic cannot accumulate Presence Credits regardless of platform policy. The exclusion is architectural. | P14 / P3 |
| 8 | Closed network creates switching costs | P9 (Trust Bridging Assertion) and P7 (Atomic Migration) enable cross-platform presence portability. A presence attestation on one platform is portable to another — eliminating the closed-network structural lock-in. | P7 / P9 |
| 9 | Advertising market distortion | P11 (Tamper-Evident Ledger) + P3 (Human Continuity Score) create an auditable presence record. Bot-driven fake engagement — the mechanism of advertiser fraud — cannot produce valid ledger entries without physical human attestation. | P11 / P3 |
| 10 | Ad blockers were circumvented | HPP operates at the presence layer, below the ad-serving layer. Bot exclusion is structural — it occurs before ad targeting, not after. Platform circumvention of presence requirements requires physical multiplication of human actors, not a software update. | P2 / P14 |
Beacon withdrawn 2007. Social plugin surveillance reversed 2014. Atlas ad server surveillance reversed 2014. User voting process abolished 2012. Do Not Track ignored indefinitely. Ad blockers circumvented 2016. Every behavioral commitment Facebook has made has been reversed within 2–7 years.
An architectural remedy is a protocol primitive embedded in the attestation layer. It cannot be reversed by a platform decision because it operates outside the platform's control surface. The H-Constant is a physical constant, not a policy setting. The hardware-bound key cannot be cloned by a software update.
The Reasonable Alternative Design (RAD) standard does not require that an alternative design be commercially deployed at time of evaluation. It requires: (1) documented technical articulation, (2) feasibility capable of evaluation through expert testimony, and (3) for pre-verdict relevance, filing dates that predate the jury's finding of design defect. HPP is not asserting legal conclusions. It is providing technical evidence — documented USPTO filing dates, formal definitions, and architectural specifications — for evaluation by counsel and expert witnesses. This section does not constitute legal advice.
| Date | Event | HPP Relevance |
|---|---|---|
| Jan 13, 2025 | MDL 3047 trial begins — KGM v. Meta et al. | Pre-filing — litigation underway |
| Feb 14, 2026 | HPP Wave 1 — 12 core applications filed | 39 days before verdict · Core RAD anchor |
| Feb 18, 2026 | HPP Wave 2 — H-Constant + Age Verification | 35 days before verdict |
| Mar 25, 2026 | KGM Verdict — Meta + YouTube liable · $6M | 2,407 cases pending · Section 230 rejected for design claims |
| Mar 27–28, 2026 | HPP Waves 3 & 4 — 29 additional applications · portfolio complete | Post-verdict · Pro-Q is enforcement architecture, not RAD anchor |
| Jun 15, 2026 | MDL 3047 Federal Trial 1 — Breathitt County (KY) | RAD documentation window closes before trial opens |
| Feb 14, 2027 | HPP NP Conversion Deadline | Priority date preservation — hard legal deadline |
W. Mark Lanier, Rachel Lanier · Lead Trial Counsel, KGM v. Meta et al. · Houston / Los Angeles
Federal bellwether trial counsel · MDL 3047 June 15, 2026 · Breathitt County, KY lead bellwether
Matthew Bergman · MDL 3047 coordination
MDL 3047 plaintiff liaison counsel
Lead outside counsel for MDL 3047. Meta stated intent to appeal KGM verdict.
Luis Li, Melissa Mills, Ariel C. Anaba, Lauren Gallo White, Samantha A. Machock. Google stated intent to appeal.
HPP's security model was submitted to USENIX Enigma 2026 before the April 1, 2026 deadline. The formal security bound was reviewed and a dimensional inconsistency (asymptotic vs. concrete query-counting terms) was identified and corrected following external cryptographic review.
The NPHT paper presents the formal security model for HPP's three formal notions and one physical assumption. All four are model-level notions in the sense of v1.2 §1.1 — formally named, quantified properties that parameterize the adversarial model. The core is the necessity theorem: no system can satisfy HPP-PRES (unforgeability) without NPHT (throughput notion), H-Constant (physical assumption), and Biometric Burn (consumption-closure notion). The construction is assumption-conditional: all security claims depending on the H-Constant are conditional on that assumption holding.
H-Constant [physical assumption — model-level notion]
→ NPHT [adversarial throughput notion — conditional on H-Constant]
→ HPP-PRES [security notion — conditional on NPHT and H-Constant]
→ Biometric Burn [consumption-closure notion — closes the construction]
Each element is necessary. Removing any one collapses the construction. Three are formal security notions; one (H-Constant) is a physical assumption that functions as a model-level notion.
Pr[A wins HPP-PRES] ≤
Adv_sig(λ,q,t)
+ q_n / 2^λ
+ q_b · ε_live
+ q_h · ε_hw
Assumption-conditional on H-Constant. Revised bound: dimensional inconsistency between asymptotic and concrete query-counting terms corrected April 2026 following external review.
H-Constant ≈ 8 joules per attestation. A physical lower-bound assumption — quantified, named, and load-bearing, but not proven within the model. Assumed in the same way that hardness assumptions (LWE, DDH, random oracle) are assumed in cryptography. Cannot be set to zero by policy or software update; grounded in the energy cost of producing a valid Biometric Liveness Gate signal on a Hardware-Bound Key device.
At 1,000,000 bot-scale attestations: ≈ 8 MJ. At 1,000,000,000: ≈ 8 GJ. Makes asymptotically free bot-scale attestation thermodynamically prohibitive — conditional on this assumption.
The NPHT adversarial game defines the throughput challenge: adversary A, given unbounded compute and parallel hardware, attempts to produce more valid attestations per unit time than a single human can produce on a single hardware-bound device within a server-authoritative Attestation Window. The NPHT necessity theorem establishes that A cannot win this game if the H-Constant assumption holds and the Biometric Liveness Gate is instantiated correctly. The sequentiality property follows: valid attestation sequences are non-parallelizable because the Biometric Burn state predicate is per-device, per-window, and non-replayable. All results in the NPHT paper are assumption-conditional on the H-Constant.
| Entity | Agile On Target LLC |
| State | North Carolina |
| Inventor | Thomas Elliott Friend (sole) |
| USPTO Customer | 224891 |
| Affiliations | None — no defendant/plaintiff ties |
| Licenses | None issued |
| Investors | None |
| NP Deadline | February 14, 2027 |
Primary priority date. Pro-A through Pro-K. App. Nos. 63/983,106–63/983,136.
Pro-L (H-Constant), Pro-P (Age Verification). App. Nos. 63/985,404 and 63/985,418.
Pro-Q, Pro-M, INV-01 through INV-15, ECO-A through ECO-H. Portfolio complete.
| Ring / Layer | Families | Primary Applications | Subject Matter |
|---|---|---|---|
| Ring 1 — Hardware | Pro-A – Pro-F | 63/983,106–63/983,121 | Secure Enclave binding, biometric pulse, attestation root, daily epoch, migration, deferred settlement |
| Ring 2 — Economics | Pro-G – Pro-K, Pro-L | 63/983,122–63/983,136, 63/985,404 | H-Constant enforcement, Presence Credits, non-transferability, anti-Sybil, session monitoring, telemetry |
| Ring 3 — Verticals | Pro-M, Pro-P, Pro-Q | 64/018,735, 63/985,418, 64/018,841 | Child safety, age verification, active/passive presence tiers (infinite scroll, autoplay) |
| INV Layer | INV-01 – INV-15 | Including 64/018,908 | 15 system invariants including P15 Endpoint-Bound Attestation (INV-15) |
| ECO Layer | ECO-A – ECO-H | — | Economic framework, Presence Credit supply, settlement, cross-platform interoperability |
All materials in this section are framed without legal conclusions, without claims of "fixing antitrust," and without advocacy framing. Architectural description only.
The Human Presence Protocol is a patented architectural framework that allows a digital system to verify, with cryptographic certainty, that a signal was produced by a physically present human being — not a bot, not an AI agent, not a script.
It works by enforcing two implementation prerequisites that bots cannot satisfy at scale without physically multiplying human actors: a Hardware-Bound Key that cannot be exported from a specific physical device, and a Biometric Liveness Gate signal that must be produced within an Attestation Window set by a server, not a client clock. These prerequisites are underpinned by three formal security notions (HPP-PRES, NPHT, Biometric Burn) and one physical assumption (H-Constant) that together constitute the Minimal Constraint Set for Verifiable Presence Unforgeability.
The result is a presence attestation — a cryptographic proof that a human being was physically present at the moment the attestation was produced. This proof cannot be faked by software alone.
The internet's current trust model assumes that signals from human beings and signals from bots are distinguishable by detection — by identifying patterns that look like bots. This assumption is increasingly false as AI improves. HPP inverts the model: instead of detecting bots, it enforces a physical cost that only humans can pay.
The H-Constant (approximately 8 joules per attestation) establishes that cost at the physical layer. It cannot be circumvented by a software update.
HPP does not surveil users. It does not evaluate what users say, share, or believe. It answers one question only: is this signal consistent with a physically present human being?
P12 (Commitment-Based Presence Proof) enables zero-knowledge verification — a platform can confirm human presence without receiving user identity. Privacy is a design requirement, not a policy aspiration.
The HPP portfolio is owned by Agile On Target LLC, a North Carolina single-member LLC. Thomas Elliott Friend is the sole inventor. The company has no affiliations to any technology platform, no investors, and no licenses issued. The portfolio was filed under USPTO Customer No. 224891.
All 42 patent families are provisional and patent pending. Non-provisional conversion must be completed by February 14, 2027 to preserve the February 14, 2026 priority dates. The portfolio is not yet issued and is subject to USPTO examination.
All claims on this site are verifiable through USPTO records, the HPP v1.2 formal construction, and the USENIX NPHT submission. The following materials are available under appropriate confidentiality arrangements for counsel, regulators, and academic reviewers.
42 provisional applications · USPTO Customer No. 224891 · Priority dates: February 14, 2026 (Wave 1, 12 applications), February 18, 2026 (Wave 2, 2 applications), March 27–28, 2026 (Waves 3–4, 29 applications).
Four notions (HPP-PRES, NPHT, H-Constant, Biometric Burn). Adversarial games. Necessity theorem. Dependency chain. Security bound. Available under NDA for regulatory and counsel review.
Submitted to USENIX Enigma 2026 before the April 1, 2026 deadline. Full paper available under NDA. Includes formal model, necessity theorem, adversarial game, and peer-review status.
HPP Antitrust Briefing v1.2. 10-condition map of Srinivasan (2019) structural mechanisms to HPP architectural responses. Available for evaluation by antitrust counsel and regulators.
Plaintiff Counsel Briefing. RAD chronology, KGM feature mapping, filing date documentation, feasibility framing. Structured for expert witness evaluation under the RAD standard.
This site releases the HPP architecture into the public domain for evaluation, verification, and academic scrutiny. All patent applications remain pending. This release does not waive any patent rights under 35 U.S.C. § 111(b).
| Entity | Agile On Target LLC |
| Jurisdiction | North Carolina |
| USPTO Customer | 224891 |